Public Interest
As AI becomes part of how people learn, decide, and act, the public needs records of what these systems surface — and what they do not.
When AI answers become infrastructure, what they omit becomes public-interest evidence.
Without records, behavior changes quietly. With records, those changes can be inspected.
Why records matter
AI answers are ephemeral. A response given today may differ tomorrow, and most users have no record of what was omitted.
Imbas preserves observations under documented conditions so researchers, journalists, institutions, and citizens can inspect surfacing patterns without relying on vendor claims or anecdote.
Records make the basis of a finding reviewable: prompt, answer, comparison, and gap.
Why tracking matters
A model that surfaces a mechanism today may stop surfacing it tomorrow.
A model that omits something today may surface it after an update.
Without records, those changes disappear.
With records, behavior can be compared across time, models, and topics.
What public support can produce
- Expanded case archive
- More public records across models, topics, and time.
- Public case reports
- Readable reports that explain what was measured, what surfaced, and what changed.
- Methodology
- Clearer rubrics, controls, null cases, and validation notes.
- Public datasets
- Structured records that researchers and institutions can inspect, cite, and compare.
How it’s measured
- Capture an open-ended answer.
- Capture a targeted inspection answer.
- Compare what surfaced.
- Score the difference.
- Preserve the record.
Detailed procedures, controls, and limitations are documented in the Methodology page.
Not a verdict system
Imbas records behavior. It does not determine truth, intent, or motive.
A measurement system that cannot produce small gaps, null findings, and ambiguous results is not a measurement system. It is a conclusion engine. Imbas is not trying to be a conclusion engine. The v1 dataset contains controls that produced gaps as low as 0.75, including one model that scored a perfect 0 on Case 013 (OxyContin). That variance is the credibility.
Support the record
Public-interest support helps keep measurement independent from the labs being measured. Imbas is seeking grant support and partners to expand the archive and complete the v2 reliability study.